Reduced Order Modeling of the In-Medium Similarity Renormalization Group

The In-Medium Similarity Renormalization Group (IMSRG) is an ab initio nonperturbative method which decouples the many-body ground state via an iterative process of continuous unitary transformations. It has been successfully used to describe finite nuclei and nuclear matter, and has been extended to valence space and equation-of-motion calculations. As the scale of IMSRG calculations increase, so too does the computational cost despite optimization and parallelization methods. A key alternative comes in the form of reduced order modeling, which decreases the overall size of the required calculation. This talk introduces two models for the IMSRG that aim to be interpretable and lightweight. The first applies a Galerkin-based operator inference emulator to reduce the dimensionality of the solution space. The second performs a perturbative expansion of the iterative process, reducing the IMSRG calculation into a matter of iterative correction rather than differential equation solving. I will demonstrate the results of these methods for an N2LO_Opt potential and discuss the future directions for emulation in large-scale many-body calculations.